:PROPERTIES:
:ID:       82127d6a-b3bb-40bf-a912-51fa5134dacc
:END:
#+title: diffusion
#+filetags: :neuralnomicon:
#+SETUPFILE: https://fniessen.github.io/org-html-themes/org/theme-readtheorg.setup
- [[id:4131b9b5-0018-4639-8285-546810516cae][AUDIO DIFFUSION]]
- [[https://arxiv.org/abs/2306.05720][Beyond Surface]] Statistics: Scene Representations in a Latent Diffusion Model
  - Stable Diffusion secretly learns 3D geometry, even without any depth data
* MULTIDIMENSION
- parent: [[id:e9be16f7-8032-4509-9aa9-7843836eacd9][domain]]
- [[https://arxiv.org/abs/2302.07400][Score-based Diffusion Models in Function Space]] function diffusion (maybe audio)
  - [[https://arxiv.org/pdf/2212.00886.pdf][Diffusion Generative]] [[https://github.com/GavinKerrigan/functional_diffusion][Models in Infinite]] Dimensions
- [[https://arxiv.org/pdf/2305.15586.pdf][Manifold Diffusion]] Fields (from texture context)
* BRAIN
- [[https://twitter.com/_akhaliq/status/1688787286807228416][Seeing through]] the Brain: Image Reconstruction of Visual Perception from Human Brain Signals
  - stimuli reconstruction by reconstructing the observed images
* NETWORK DIFFUSION
- [[https://twitter.com/_akhaliq/status/1760136914861019645][Neural]] Network Diffusion
  - utilizing an autoencoder to extracts latent representations of a subset of the network parameters
  - then trained to synthesize these
  - improved performance over trained networks